Gene expression signatures define novel oncogenic pathways
in T cell acute lymphoblastic leukemia
Adolfo A. Ferrando,1
Donna S. Neuberg,2
Mignon L. Loh,4,8
Susana C. Raimondi,5
Fred G. Behm,5
James R. Downing,5
D. Gary Gilliland,4
Eric S. Lander,3
Todd R. Golub,1,3
and A. Thomas Look1,7
1Department of Pediatric Oncology 2Department of Biostatistical Science
Dana-Farber Cancer Institute and Harvard Medical School, Boston, Massachusetts 02115
3Whitehead Institute/Massachusetts Institute of Technology Center for Genome Research, Cambridge, Massachusetts 02142 4Department of Medicine, Brigham and Women’s Hospital, Boston, Massachusetts 02115
5Department of Pathology
6Department of Hematology/Oncology
St. Jude Children’s Research Hospital, Memphis, Tennessee 38105
8Present address: Department of Pediatrics, University of California, San Francisco, California 94143
9Present address: Millennium Pharmaceuticals, Inc., Predictive Medicine Group, Cambridge, Massachusetts 02139
Human T cell leukemias can arise from oncogenes activated by specific chromosomal translocations involving the T cell
receptor genes. Here we show that five different T cell oncogenes (HOX11, TAL1, LYL1, LMO1, and LMO2) are often
aberrantly expressed in the absence of chromosomal abnormalities. Using oligonucleotide microarrays, we identified
several gene expression signatures that were indicative of leukemic arrest at specific stages of normal thymocyte
signature (pro-T), HOX11ⴙ
(early cortical thymocyte), and TAL1ⴙ
(late cortical thymocyte). Hierarchical clustering
analysis of gene expression signatures grouped samples according to their shared oncogenic pathways and identified
HOX11L2 activation as a novel event in T cell leukemogenesis. These findings have clinical importance, since HOX11
activation is significantly associated with a favorable prognosis, while expression of TAL1, LYL1, or, surprisingly, HOX11L2
confers a much worse response to treatment. Our results illustrate the power of gene expression profiles to elucidate
transformation pathways relevant to human leukemia.
2000; Rivera et al., 1991). Securing further advances in treatment
outcome will likely prove difficult without improved knowledge
of the factors that contribute to the malignant behavior of
trans-T cell acute lymphoblastic leukemia (trans-T-ALL) is a malignant
dis-formed thymocytes. Unfortunately, most of the clinical and
labo-ease of thymocytes, accounting for 10%–15% of pediatric and
ratory features that guide therapy for B cell precursor ALL are
25% of adult ALL cases. Patients with T-ALL tend to present
only marginally useful in T-ALL (Pullen et al., 1999).
with very high circulating blast cell counts, mediastinal masses,
Current understanding of the molecular basis of T-ALL has
and central nervous system involvement. The prognosis of
come largely from analysis of recurrent chromosomal
transloca-T-ALL in children and adolescents has improved in recent years
tions and intrachromosomal rearrangements. These
abnormali-due to intensified therapies, with 5 year relapse-free survival
ties typically juxtapose strong promoter and enhancer elements
rates now in the range of 60%–75% (Pui and Evans, 1998;
Silverman et al., 2001; Chessells et al., 1995; Schrappe et al.,
responsible for high levels of expression of T cell receptor genes
S I G N I F I C A N C E
Careful analysis of clonal chromosomal abnormalities in leukemic blast cells has been a catalyst for the development of new diagnostic and therapeutic strategies. However, this line of research has had a much greater impact on the B lineage leukemias than on T cell acute lymphoblastic leukemia (T-ALL), whose pathogenesis and molecular subtypes remain largely undefined. Using a combination of DNA microarray and RT-PCR methods to analyze clinical T-ALL samples, we obtained results in support of our central hypothesis that aberrant activation of certain key transcription factor genes, often in the absence of chromosomal rearrangements, is the principal transforming event in this disease. These developmentally important molecules are shown to drive a limited number of oncogenic pathways with prognostic significance. The ability to classify T-ALL according to shared pathways of leukemic transforma-tion has important implicatransforma-tions for future research. It provides a conceptual framework in which to identify specific genes that determine treatment responsiveness and should foster the development of successful new therapies directed to critical molecules in pathologic transcriptional cascades. The hypothesis-driven approach to microarray analysis described in this report may also be useful for the study of other types of human cancers.
next to a small number of developmentally important transcrip-
mosome band 1p32 that are known to cause aberrant
expres-sion of TAL1 in T-ALL. However, 9 (31%) of the 29 cases with
tion factor genes, including HOX11/TLX1, TAL1/SCL, TAL2,
LYL1, BHLHB1, LMO1, and LMO2 (Finger et al., 1989; Xia et
increased expression of TAL1 (Figure 1B) had the Tal1d variant,
which results from a small deletion next to the TAL1 locus (Aplan
al., 1991; Mellentin et al., 1989; Wang et al., 2000; McGuire et
al., 1989; Royer-Pokora et al., 1991; Dube et al., 1991; Hatano
et al., 1990; Bernard et al., 1991). Thus, the majority of cases
with high levels of TAL1 oncogene expression (22/29) lacked
et al., 1991; Fitzgerald et al., 1991; Kennedy et al., 1991; Lu et
al., 1991; Brown et al., 1990; Aplan et al., 1991, 1992; Baer,
cytogenetic or molecular evidence of rearrangements affecting
the TAL1 locus, in agreement with results of an earlier study
1993; Begley et al., 1989; Chen et al., 1990; Greenberg et al.,
1990; Boehm et al., 1991), resulting in their aberrant expression
(Bash et al., 1995).
Thirteen (22%) of the 59 cases were classified as LYL1⫹
in developing thymocytes. Although the oncogenicity of these
proteins is well established (Larson et al., 1994; Neale et al.,
(Figure 1C) on the basis of LYL1 expression levels that exceeded
the mean value in normal thymocytes by more than 5-fold.
1995; McGuire et al., 1992; Chervinsky et al., 1999; Condorelli
et al., 1996; Kelliher et al., 1996; Hawley et al., 1997), under-
Increased expression of this oncogene was not associated with
cytogenetic abnormalities affecting the LYL1 locus (19p13),
con-standing of the downstream transcriptional programs that
gen-erate and maintain the T-ALL phenotype remains limited. Further
sistent with the paucity of reports on LYL1 activation by
chromo-somal translocation. Thus, other mechanisms appear
responsi-improvement of risk-based treatment strategies and the
devel-opment of effective new drugs for T-ALL will depend on fresh
ble for the aberrant expression of LYL1 in thymic leukemias.
Finally, TAL2 and BHLHB1 were expressed at high levels in a
insights into the molecular pathways usurped by HOX11, TAL1,
and other oncoproteins in developing thymocytes.
single case each (Figure 1C). Again, cytogenetic analysis failed
to reveal locus-specific translocations associated with the
ex-pression of these oncogenes. Some cases overexpressed more
than one of the closely related bHLH T cell oncogenes: six
overexpressed both TAL1 and LYL1, and one overexpressed
Oncogenic transcription factor expression in T-ALL
HOX11, an orphan homeobox gene essential for splenic devel-
TAL1 and BHLHB1.
Analysis of the LIM-only domain genes LMO1 and LMO2
opment (Roberts et al., 1994; Dear et al., 1995), is activated
in a subset of T-ALL cases bearing the t(10;14)(q24;q11) or
(Figure 1D) showed an absence of significant expression of
either LMO gene in the HOX11⫹
overexpres-t(7;10)(q35;q24), each of which places HOX11 under the control
of strong enhancers embedded in the T cell receptor loci. Acting
sion of one of these genes was observed in most samples
overexpressing TAL1, and high levels of LMO2, but not LMO1,
on the hypothesis that HOX11 might be aberrantly expressed
in cases other than those harboring locus-specific transloca-
were found in the LYL1⫹
samples. These results are consistent
with biochemical and transgenic animal model studies showing
tions, we used quantitative real-time reverse transcriptase PCR
(RT-PCR) to analyze HOX11 expression, detecting high levels
that LMO proteins form heterocomplexes and act in concert
with TAL1 and possibly other bHLH proteins in T-ALL
(Valge-of this oncogene in 8 (Valge-of 59 pediatric T-ALL samples (Figure 1A).
Four of the HOX11-positive cases had cytogenetic abnor-
Archer et al., 1994; Wadman et al., 1994; Larson et al., 1996;
Chervinsky et al., 1999; Herblot et al., 2000; Wadman et al.,
malities involving the HOX11 locus in chromosome band
10q24: two t(10;14)(q24;q11.2), one t(7;10)(q35;q24), and one
1997). Ten of the 59 cases (samples 48–51, 53–57, and 59,
Figure 1) did not express abnormal levels of any of the
transcrip-del(10)(q24q26). Reduced levels of HOX11 expression (100–
1000 times lower than in the eight HOX11⫹
samples) were de-
tion factor genes described above, raising the possibility of
thymocyte transformation via alternative oncogenic
mecha-tected in four additional samples, two of which also had
cytoge-netic abnormalities of band 10q24 [t(10;14) and add(10)(q24)].
The remaining cases and eight normal control thymus samples
showed only background levels of HOX11 expression, near the
Gene expression profiles and their biologic correlates
T cell development is a tightly regulated multistep process that
limit of detection with this technique. Thus, by using quantitative
RT-PCR analysis of the HOX11 gene, we identified a substantial
involves the intrathymic differentiation, proliferation, and
selec-tion of T cell precursors (Murre, 2000; Rodewald and Fehling,
proportion of HOX11⫹
cases that express high levels of the
oncogene while lacking cytogenetically detectable alterations
1998). Leukemic thymocytes retain many of the biologic features
of normal T cell precursor subpopulations, as illustrated by
of the 10q24 region.
Prominent among T cell oncoproteins are members of the
shared patterns of cell surface protein expression (Reinherz and
Schlossman, 1980; Bene et al., 1995). We therefore postulated
basic helix-loop-helix (bHLH) family of transcription factors:
TAL1, TAL2, LYL1, and the recently described BHLHB1 protein
that HOX11, TAL1, and LYL1 overexpression might directly or
indirectly interfere with transcriptional networks that normally
(Bernard et al., 1990; Finger et al., 1989; Xia et al., 1991;
Mellen-tin et al., 1989; Wang et al., 2000). These transcriptional regula-
regulate thymocyte proliferation, differentiation, and survival
during T cell development (Look, 1997). To test this hypothesis,
tors are believed to act through a common mechanism involving
dominant negative interference with the activities of the E47
we used oligonucleotide microarrays (Affymetrix, HU6800) to
analyze the global patterns of gene expression in 39 of the
and E12 variants of E2A transcription factors (Begley and Green,
1999; Wang et al., 2000; Miyamoto et al., 1996; Hsu et al., 1994;
T-ALL samples with sufficient RNA for these studies (Golub et
al., 1999). The microarray data are available in their entirety as
Park and Sun, 1998), whose homozygous inactivation leads
to T cell tumors in mice (Bain et al., 1997; Yan et al., 1997).
Supplemental Data at http://www.genome.wi.mit.edu/mpr and
Quantitative RT-PCR analysis revealed increased levels of TAL1
mRNA in 29 (49%) of the 59 cases (Figure 1B). None of these
We first asked whether the results obtained from microarray
hybridizations agreed with the quantitative RT-PCR findings
samples harbored any of the recurrent translocations of
chro-Figure 1. Quantitative RT-PCR analysis of oncogenic transcription factor genes in pediatric T-ALL samples and normal thymus controls
A: HOX11 expression. Samples with expression levels⬎1 ⫻ 105mRNA copies/100 ng of RNA (dotted line) were considered HOX11⫹. Cases with abnormalities of chromosome band 10q24 are indicated with arrows.
B: TAL1 expression. Crosshatched bars indicate samples with the Tal1d variant band in 1p32, resulting from deletion of a 90 kb genomic DNA fragment adjacent to the TAL1 locus. Samples showing TAL1 expression levels above that detected in TAL1d⫹samples were considered TAL1⫹(dotted line). C: Expression of LYL1 (yellow bars) and other bHLH transcription factor genes (TAL2 and BHLHB1) (green bars). The threshold level for LYL1 positivity (dotted line) corresponds to five times the mean level of expression in normal thymus control samples.
D: Expression of LMO1 (purple bars) and LMO2 (orange bars). The threshold level for LMO1 and LMO2 positivity (dotted line) corresponds to five times the mean level of LMO2 expression in normal thymus control samples. Oval symbols in panels A, B, and C indicate cases included in the microarray analysis.
presented in Figure 1. Normalized microarray results, plotted
sion of the CD1 (A-E family members), LAR, and CD10 genes
in a pattern resembling that of normal cells undergoing the early
as increasing intensities of red (positive) or blue (negative)
rela-tive to the mean value, are shown in Figure 2. The first row of
cortical stage of thymocyte differentiation (Terstappen et al.,
1992; Terszowski et al., 2001; Rodewald and Fehling, 1998).
colored squares in each of the three panels depicts the
expres-sion levels of HOX11 (top), TAL1 (middle), or LYL1 (bottom)
Many of the genes associated with HOX11 expression are
in-volved in cell growth and proliferation. These include adenosine
among the 27 cases independently expressing one of the three
major oncogenes. The cases are arranged in the same order
deaminase (target of pentostatin, fludarabine, and
2-chloro-deoxyadenosine), DNA topoisomerase (target of the
anthracy-used to display the quantitative RT-PCR data. There was
re-markable overall agreement between gene expression values
clines and epipodophyllotoxins), dihydrofolate reductase (target
of methotrexate), hypoxanthine phosphoribosyltransferase 1
obtained by these two methods.
We next surveyed genes considered to be “nearest neigh-
(modifier of the effect of antimetabolite therapy), and thymidylate
synthetase (target of fluoropyrimidines and other novel
folate-bors” (Golub et al., 1999) of HOX11, TAL1, and LYL1, based
on the close agreement of their expression profiles. Analysis of
based inhibitors). The gene products DNA polymerase epsilon,
cyclin A, Tax1 binding protein, and replication protein A1 all
the resultant gene expression signatures (Figure 2) revealed a
striking concordance with recognized stages of normal thymo-
have prominent roles in cell proliferation. These findings are
consistent with data showing that HOX11 can both immortalize
cyte development (Figure 3). Similar findings have been reported
for B lineage tumors studied with cDNA microarray technology
hematopoietic progenitors (Keller et al., 1998) and interact
di-rectly with cell cycle regulatory proteins (Kawabe et al., 1997).
(Allzadeh et al., 2000). HOX11⫹
cases showed increased
expres-For example, high levels of HOX11 correlated with increased
expression of MYC and the proapoptotic glucocorticoid
recep-tor gene. TAL1 overexpression was associated with the
upregu-lation of proto-oncogenes such as CBFA2 (AML1) and the
MYB-related gene MYBL2, receptor genes such as IL8R and CSFR1,
and the antiapoptotic gene BCL2A1. Finally, LYL1 positivity was
related to higher expression levels of the MYCN, LMO2, and
PLZF proto-oncogenes, as well as the antiapoptotic gene BCL2.
Most antineoplastic drugs are thought to act through the
mito-chondrial apoptotic machinery, and their cytostatic effects are
inhibited by BCL2 and its related prosurvival family members
(Reed, 1995). Thus, the upregulation of BCL2 and BCL2A1 in
LYL1- and TAL1-overexpressing cases may explain their relative
resistance to chemotherapy (see Figure 5), while the exquisite
Figure 3. Correlation of gene expression profiles in LYL1⫹, HOX11⫹, and TAL1⫹
responsiveness of HOX11⫹
cases could partly reflect the
down-T-ALL samples with recognized stages of thymocyte differentiation
regulation of survival factors in early cortical stage thymocytes,
Cell surface markers normally associated with each developmental stage
most of which are targeted for “death by neglect.”
(Murre, 2000) are indicated in black, with corresponding microarray findings shown in red. The most immature T cell precursors express CD34 but not
CD4, CD8, or CD3. As these cells mature, they lose CD34 expression while
Hierarchical clustering of T-ALL cases based on genegaining CD4 and then CD8, becoming double-positive thymocytes. Early
double-positive cells initially express CD1 and CD10 (early cortical
thymo-Although helping to identify the sets of genes coordinately
ex-cytes). As they finish rearranging their T cell receptor genes, 95% of these
cells fail to express a functional receptor and are ablated through a death-
pressed with HOX11, TAL1, and LYL1, the nearest neighbor
by-neglect mechanism. Thymocytes with functional T cell receptors gain
analysis depicted in Figure 2 provided little useful informationCD3 expression (late cortical thymocytes) and undergo both positive and
about the 10 cases that lacked discernible expression of these
negative selection. Cells surviving this process proceed through a final step
oncogenes (designated Other in Figure 4) or the two additional
of differentiation in which they downregulate the expression of either CD4
or CD8 to become mature single-positive T cells.
cases expressing both LYL1 and TAL1 (Mixed). To gain insight
in the molecular characteristics of these poorly understood
cases, we generated hierarchical clusters based on the 72 genes
whose expression patterns best distinguished between each
Because the drugs used to treat human leukemias are more
group of HOX11⫹
, and other cases in pairwise
active in proliferating cells, these findings may explain in part
comparisons (as defined by the permutation distribution of the
the better prognosis of patients with HOX11⫹
T-ALL (see Fig-
maximum t statistic, P
As shown in Figure 4, the HOX11⫹
, and LYL⫹
sam-By contrast, the expression pattern associated with TAL1
ples detected by RT-PCR are grouped together within major
expression appeared to reflect the late cortical stage of thymo-
branches of the dendrogram. The branch containing the HOX11
cyte differentiation, as indicated by the upregulation of LCK,
samples (H) comprises two subgroups, one containing most of
TCRA, TCRB, CD2, CD6, and CD3E (Terstappen et al., 1992;
the HOX11 RT-PCR-positive cases (H1) and the other consisting
Rodewald and Fehling, 1998). High levels of LYL1 expression
primarily of HOX11-like samples that lacked HOX11 expression
were associated with an undifferentiated thymocyte phenotype
by RT-PCR (H2). Surface immunophenotyping indicated that
characterized by increased expression of the early hematopoi-
these subgroups had related but distinct immunophenotypes
etic marker gene CD34, the cell adhesion gene L-selectin (SELL),
(Table 1). True HOX11 samples were primarily CD1⫹
the antiapoptotic gene BCL2, and LSP1, which encodes the
, and CD3⫺
(early cortical thymocytes), while the
lymphocyte-specific protein 1 (Pilarski et al., 1991; Galy et al.,
HOX11-like samples were primarily CD1⫹/⫺
1993; Ma et al., 1995; Palker et al., 1998). These results suggest
, and CD3⫹
(early cortical thymocytes with acquired CD3
that T cell oncogenes specifically interfere with transcriptional
surface expression). Similarly, the central cluster of TAL1⫹
sam-programs controlling thymocyte development, leading to stage-
ples contained two subgroups, a larger one comprising most
specific developmental arrest.
of the true TAL1⫹
samples (T1) and a second consisting mainly
The three molecularly distinct subtypes of T-ALL also
of TAL1-like samples (T2). A third, smaller branch (M) emerged
showed specific associations with known proto-oncogenes, as
from the hierarchical analysis, and was characterized by a global
pattern of increased expression of many of the genes that
distin-well as genes involved in programmed cell death (Figure 2).
Figure 2. HOX11⫹, TAL1⫹, and LYL1⫹nearest neighbor analysis
Each row of squares shows the expression pattern of a particular gene selected by nearest neighbor analysis (Golub et al., 1999), while each column represents 1 of the 27 samples positive for HOX11, TAL1, or LYL1 by RT-PCR (see Figure 1). The genes depicted were chosen from the top 200 nearest neighbors of each major oncogene (boldface type) on the basis of their potential functional relevance and then were grouped according to their involvement in T cell differentiation, apoptosis, cell proliferation, or chemotherapy response. Expression levels for each gene were normalized across the samples; levels greater than or less than the mean (by as much as three standard deviations) are shown in shades of red or blue, respectively. Numbers at the bottom correspond to the numbers of the samples in Figure 1. For a complete list of gene names, accession numbers, and raw expression values, see Supplemental Data at http://www.genome.wi.mit.edu/mpr and http://www.cancercell.org/cgi/content/full/1/1/75/DC1.
Figure 4. Hierarchical cluster analysis of gene expression data
Expression profiles of 72 genes were selected by permutation test analysis as those best distinguishing among 39 HOX11⫹, TAL1⫹, LYL1⫹cases and unclassified samples (Other). The dendrogram (top) shows the relatedness of gene expression among samples and is color coded according to the quantitative RT-PCR category of each sample (see Figure 1). Clinical outcome data are reported as horizontal bars with open boxes representing survivors and dark boxes deceased patients. Cytogenetic and molecular abnormalities are indicated by discrete symbols defined at the bottom of the figure. Each column represents a T-ALL mRNA sample and each row a gene on the microarray. Genes are grouped into four consecutive categories: higher in HOX11⫹than in TAL1⫹, LYL1⫹, or Others; higher in TAL1⫹than in HOX11⫹, LYL1⫹, or Others; higher in LYL1⫹than in HOX11⫹, TAL1⫹, or Others; and finally higher in Others than in HOX11⫹, TAL1⫹, or LYL1⫹; and are listed within each category in order from lowest to highest P value. A complete list of genes and P values is available as Supplemental Data at http://www.genome.wi.mit.edu/mpr and http://www.cancercell.org/cgi/content/full/1/1/75/DC1. Gene expression values are normalized and color coded, as indicated by the scale beneath the graph. Major branches in the dendrogram are designated by the first letter of the dominant oncogene (e.g., H, H1, H2, for HOX11).
guished among the other three groups. Interestingly, two of
Finally, the LYL⫹
cluster (L) included two branches. One
contained three of the true LYL⫹
samples (L1), including the
these three cases had the t(11;19)(q23;p13.3), which produces
the MLL-ENL fusion gene (Rubnitz et al., 1999a), while the re-
only T cell sample in this series with the FLT-3 internal tandem
duplication, which is often identified in acute myeloid leukemias
maining case had a normal karyotype. MLL-ENL RT-PCR
analy-sis, performed in 59 samples, revealed the MLL-ENL fusion
(Nakao et al., 1996; Yokota et al., 1997). These leukemias also
expressed high levels of CD34 as well as myeloid markers,
transcript in only three cases, all in the M cluster, including the
case with a normal karyotype. This result illustrates the power of
consistent with differentiation arrest in the early stages of T cell
development, when T progenitor cells are migrating from the
DNA microarray analysis to group samples according to specific
Table 1. Cell surface antigen expression among T-ALL samples DPa DNb MYc CD34⫹ CD10⫹ CD1⫹ CD4⫹CD8⫹ CD4⫺CD8⫺ CD3⫹ CD13⫹/33⫹ Cluster H 43 78 75 78 0 36 7 n⫽ 14 HOX11⫹ 37 62 100 62 0 0 0 In cluster H n⫽ 8 HOX11⫺ 50 100 50 100 0 83 16 In cluster H n⫽ 6 Cluster T 71 27 46 73 13 66 0 n⫽ 14 Cluster T1 78 20 40 70 20 90 0 n⫽ 9 Cluster T2 60 40 60 80 0 20 0 n⫽ 5 Cluster L 100 33 0 0 83 33 100 n⫽ 6
The values are percentages of positive samples. aDouble-positive thymocytes.
bDouble-negative thymocytes. cMyeloid lineage.
sample and two samples with simultaneous expression
group of leukemias. HOX11L2, on the other hand, was
ex-pressed at high levels (
copies per 100 ng of RNA) in six
of TAL1 and LYL1 by quantitative RT-PCR. Indications of
multistep mutational pathways emerged when samples were
of the T-ALL samples (29, 48, 49, 51, 54, and 55, as numbered
in Figure 1), but was undetectable in normal thymus and the
analyzed by quantitative DNA PCR for deletions of
P16/INK4A-P14/ARF (Drexler, 1998; Okuda et al., 1995). In our series of
other T-ALL samples. Three of the six HOX11L2⫹
triangles in Figure 4) had sufficient RNA for microarray analysis.
T-ALL cases, homozygous deletions of this gene were found
in most samples in clusters H and T (Figure 4), which included
Their location in the HOX11-related “H2 cluster” of the
hierarchi-cal dendrogram confirms our hypothesis that cases with gene
cases as well as cases with similar overall
patterns of gene expression. Homozygous deletion of P16/
expression signatures resembling HOX11⫹
cases might be
transformed through the effects of highly related oncogenes
INK4A-P14/ARF was not detected in 2 of the three MLL-ENL⫹
cases in the M cluster, nor in any of the cases grouped into
operating through similar oncogenic pathways.
Statistical analysis to identify the genes that were
differen-cluster L, which comprise LYL1⫹
cases as well as mixed cases
expressing both TAL1 and LYL1 (Figure 4). The exclusive pres-
tially expressed in HOX11⫹
increased expression (Permax P value
⬍ 0.30, see Experimental
ence of cytogenetic features such as the 5q- and 13q-deletions
within the L2 subgroup (Figure 4) further attests to the ability
Procedures) of HOX11 itself and eight additional genes in
cases (Fuse binding protein 2 [FBP2; U69126],
of our hierarchical clustering approach to group samples with
common mechanisms of transformation and suggests that tu-
DXS9879E [ITBA2; X92896], H2AZ histone [H2AZ; M37583],
mor suppressor genes in the 5q and 13q regions are inactivated
as part of a distinct oncogenic pathway that gives rise to these
18.104.22.168; X01677], SW1/SNF complex 155 kDa subunit
[BAF155; U66615], FYN binding protein [FYB; U93049],
␣ 1 [PSMA1; M64992] and tubulin ␤ 5 [V00599]).
Oncogene discovery through microarray expression
None of the genes on the microarray were expressed at
signifi-analysis: HOX11L2 is activated in T-ALL samples
cantly higher levels in HOX11L2⫹
cases. Thus, despite the
The observation that T-ALL leukemia cases with MLL-ENL re-
marked similarities in gene expression profiles between HOX11⫹
arrangement and recurrent cytogenetic abnormalities were
cases are distinguished by
in-grouped together in our hierarchical clustering analysis illus-
creased expression of genes involved in signal transduction
trates the ability of gene expression profiling to identify cases
and the chromatin-mediated control of gene expression (see
with common mechanisms of transformation. Results of this
analysis also suggested that cases without defined oncogene
activation that clustered with the HOX11, TAL1, or LYL1 samples
Oncogene activation and gene expression signatures
have prognostic relevance
likely harbor related but as yet unidentified oncogenes. To test
this hypothesis, we used quantitative RT-PCR to analyze the
To assess the prognostic significance of these findings, we first
analyzed the survival of 58 eligible patients from the 59 whose
expression of HOX11L1 and HOX11L2, two homeobox genes
that are not included in the Affymetrix 6800 microarray but are
leukemic cells were analyzed for HOX11, HOX11L2, TAL1, or
LYL1 expression by quantitative RT-PCR. Preliminary
compari-functionally and structurally related to HOX11. HOX11L1 was
expressed at comparable low levels in both normal thymus and
son of the Kaplan-Meier plots showed no significant difference
between the TAL1⫹
groups, prompting us to
com-T-ALL samples, indicating that it was not overexpressed in this
achieve complete remission, was noted only in patients with
overexpression of TAL1 or LYL1.
Similar results were obtained when the Kaplan-Meier
analy-sis focused on the major groups defined by hierarchical
cluster-ing of gene expression signatures (Figure 5B): probability of
survival at 5 years was 92%
⫾ 8% for the HOX11⫹
⫾ 19% and 33% ⫾ 19% for the TAL1⫹
⫽ 0.03). None of the three patients in the small cluster
containing the MLL-ENL cases have died. Although based on
a small number of samples, this result agrees with a previous
report suggesting that MLL-ENL translocations in T-ALL may
not carry the dire prognosis associated with related
transloca-tions in infants and older children with an early B lineage ALL
immunophenotype (Behm et al., 1996; Rubnitz et al., 1999b).
The dramatically different clinical courses of T-ALL patients
treated with the same intensive multidrug regimens support the
interpretation that ALL arising in thymic lymphocytes comprises
several biologically distinct diseases. This variability likely
re-flects a molecular heterogeneity that has not been appreciated
from characterization of leukemic T cells by conventional
meth-ods, as recently demonstrated in a microarray study for diffuse
large B cell lymphoma (Allzadeh et al., 2000). Thus, gene
expres-sion analysis using oligonucleotide or cDNA microarrays offers
a novel tool for delineating molecular pathways that drive the
malignant transformation of developing thymocytes.
The results reported here identify previously unrecognized
molecular subtypes of T-ALL and link the activation of particular
oncogenes to defined stages of normal thymocyte
develop-ment. Highly favorable clinical outcomes were observed for
patients in the HOX11⫹
cluster, whose cell samples showed a
pattern of gene expression resembling that of early cortical
thymocytes. The better therapeutic responsiveness of this
sub-group may be explained by several distinctive features of
lymphoblasts, including the expression of genes
asso-ciated with increased cell proliferative activity and the lack of
expression of BCL2 and related antiapoptotic genes. Apoptosis
is a major regulatory mechanism during normal T cell
develop-Figure 5. Clinically important T-ALL subgroups identified by gene expression
ment, eliminating the more than 90% of normal corticalthymo-profiling
cytes unable to express functional T cell receptors (Vacchio et
A: Kaplan-Meier plots of overall survival among patients with a HOX11⫹,
al., 1998). Thus, HOX11⫹
lymphoblasts appear to be arrestedHOX11L2⫹, bHLH⫹, or unclassified (Other) gene expression signature by
RT-at a stage of thymocyte development thRT-at is especially
respon-PCR analysis (TAL1⫹, TAL2⫹, bHLHB1⫹, and LYL1⫹samples combined as
sive to drug-induced programmed cell death.
B: Kaplan-Meier plots of overall survival for subgroups recognized by hierar-
Less favorable outcomes were observed in subgroupsde-chical clustering of DNA microarray data, which subdivided the samples
fined by gene expression profiles characteristic of TAL1⫹
orinto four main clusters (H, T, M, and L; see Figure 4). Tick marks on the curves
samples, which resemble late cortical and early pro-T
represent surviving patients.
thymocytes, respectively. Drug resistance in LYL⫹
be explained by the fact that early double-negative pro-T cells
express high levels of BCL2 and show increased resistance to
apoptosis (Veis et al., 1993). TAL1⫹
cells appear to upregulate
bine these two cohorts (bHLH⫹
) for further analysis. As shown
BCL2A1 (also known as BFL1) and other antiapoptotic
mole-in Figure 5A, constitutive expression of HOX11 was associated
cules normally induced by signaling through the TCR in late
with a favorable prognosis: probability of survival at 5 years was
cortical thymocytes (Tomayko et al., 1999), suggesting different
⫾ 0% standard error (SE), compared with 30% ⫾ 24%
mechanisms of treatment resistance in TAL1⫹
for the HOX11L2 group and 51%
⫾ 9% for the bHLH⫹
We find it surprising that LYL1 and TAL1 overexpression is
⫽ 0.02 by log-rank analysis, comparing HOX11⫹
with all other
associated with maturational arrest at opposite ends of the
patients). Patients whose leukemic cells lacked expression of
thymocyte developmental spectrum, despite structural and
bio-HOX11, HOX11L2, or bHLH oncogenes had essentially the
chemical data suggesting that these two proteins might act
same probability of 5 year survival as the latter two groups (P
(Murre, 2000; Baer, 1993; Miyamoto et al., 1996). This may
would stress, however, that only a small subset of the genes
comprising each gene expression signature are likely to be
di-reflect differences in the stages of thymocyte differentiation
at which these oncogenes are activated. Alternatively, the key
rectly regulated by the oncogenic transcription factors
them-selves. Since many of the specifically expressed genes appear
transformation event involving bHLH oncogenes may occur
early in the CD4⫺
cell stage. In this model, TAL1 may
to reflect a specific stage of T cell developmental arrest, it will
be important to compare gene expression profiles in leukemic
abrogate the normal E2A-induced arrest of further differentiation
(Engel et al., 2001) more effectively than LYL1, leading to leuke-
T lymphoblasts versus subsets of normal thymocytes at different
stages of differentiation to identify transcriptional programs that
mias that resemble more mature CD4⫹
are directly linked to leukemic transformation.
Our studies indicate that wider application of gene
expres-Our microarray studies of leukemic thymocytes revealed
distinctive gene expression signatures that are strongly associ-
sion profiling in T-All would help to identify therapeutically
rele-vant diagnostic subgroups. It may also be possible, given
suffi-ated with specific oncogenic transcription factors. In some
in-stances, closely related signatures were found in samples lack-
cient numbers of patients, to identify signal transduction
pathways that are vital to the proliferation and survival of
individ-ing activation of known T-ALL oncogenes, leadindivid-ing us to predict
alternative oncogenic transcription factors that could initiate
ual subgroups, making proteins within these pathways attractive
targets for new therapeutic approaches.
similar patterns of gene expression. In experiments based on
this hypothesis, we identified HOX11L2 overexpression as an
oncogenic event in HOX11-negative samples that exhibited the
gene expression signature associated with bona fide
expressing cases. HOX11L2 is an orphan homeobox factor verySamples of cryopreserved lymphoblasts from 59 children and young adults
similar to HOX11, and has been shown to be essential for thewith T-ALL, treated in Total Therapy studies XI–XIII at St. Jude Children’s
normal development of the ventral medullary respiratory center,Research Hospital (TN), were obtained with informed consent at the time of diagnosis, before any chemotherapy was given. The median age of the
in that its deficiency in mice leads to a respiratory failure
resem-patients was 9.3 years (range 0.5–18.8), the male to female ratio was 3.0,
bling congenital central hypoventilation syndrome in humans
and the leukocyte count at diagnosis was 2,300–917,000 per mm3(median,
(Shirasawa et al., 2000). The recent report by Bernard et al.
164,000). Mean lymphoblast percentage in the samples analyzed was 91%⫾
(2001) of a novel cryptic recurrent translocation t(5;14)(q35;q32)10% SD. Six patients had CNS disease at presentation, and mediastinal
in T-ALL resulting in aberrant HOX11L2 expression reinforcesmasses were present in 36. One case with less than one year of followup
the role of this homeobox factor as a T-ALL oncogene.was excluded from survival analysis. Lymphoid cells were also obtained (with informed consent) from normal thymic tissue removed at the time of
Given the marked similarity in gene expression profiles
cases, it is surprising that these
two groups of patients have such different treatment outcomes,
DNA and RNA preparation
and additional patients will need to be studied to confirm this
RNA was prepared from cryopreserved lymphoblasts with RNAqueous
re-result. Of the eight genes that were more highly expressed byagents (Ambion) according to the manufacturer’s instructions and
leukemias, FBP2, BAF155, and FYB encode regulatorytated spectrophotometrically. The quality of the purified RNA was assessed by visualization of 18S and 28S RNA bands under ultraviolet light after
proteins that might provide insight into the dissimilar clinical
electrophoresis through denaturing agarose gels and staining with ethidium
responses. The far upstream binding protein 2 (FBP2) regulates
bromide. Genomic DNA from each sample was extracted with a commercial
alternative mRNA splicing through binding to intronic splicing
kit (GENTRA) following the manufacturer’s instructions,
spectrophotometri-enhancer sequences (Min et al., 1997), while BRG1-associated
cally quantified, and stored at⫺20⬚C until analysis.
factor 155 (BAF155) is the human homolog of the yeast protein
SWI3, a component of the SWI/SNF complex that regulatesPrimers and probes
gene expression through chromatin remodeling. Two compo-Primers and probes were designed with the assistance of the computer program Primer Express (Perkin-Elmer Applied Biosystems) and with flanking
nents of the corresponding mammalian complex, BAF47/SNF5
intron-exon boundaries to prevent amplification from any residual genomic
(Roberts et al., 2000; Versteege et al., 1998) and BRG1 (Wong
DNA, while avoiding areas involved in the generation of alternative spliced
et al., 2000), are known to be tumor suppressors, and BAF155
mRNAs. In the case of TAL2, BHLHB1, HOX11L1, and HOX11L2, which
together with BRG1 has been shown to interact with cyclin E,
lacked suitable intron-exon boundaries for primer-probe design, the amount
with BRG1 specifically causing cell growth arrest (Shanahan etof residual genomic DNA in each sample was determined by simultaneous
al., 1999). FYN binding protein (FYB) is an important positivequantitation of these genes on RNA specimens in the presence and absence of reverse transcriptase. GAPDH FW: 5⬘-GAAGGTGAAGGTCGGAGT-3⬘,
regulator of T cell activation and couples TCR signals to integrin
GAPDH RV: 5⬘-GAAGATGGTGATGGGATTTC-3⬘, GAPDG Probe:
5⬘-VIC-activation and adhesion (Geng et al., 2001; Griffiths et al., 2001;
CAAGCTTCCCGTTCTCAGCC-TAMRA-3⬘, TAL1 FW: 5⬘-GAAGAGGAGA
Peterson et al., 2001).
CCTTCCCCCT-3⬘, TAL1 RV: 5⬘-GGTGAAGATACGCCGCACA-3⬘, TAL1
In contrast to the inclusive microarray analysis employed
Probe: 5⬘-FAM-TGAGATGGAGATTACTGATGGTCCCCA-TAMRA-3⬘, TAL2
by Allzadeh and coworkers to characterize subgroups of B cell
FW: 5⬘-GCCTGCAACAAACGGGAGT-3⬘, TAL2 RV: 5⬘-AGAGTTCTGTCCTC
lymphoma (Allzadeh et al., 2000), we chose to focus genes thatCAGGCCT-3⬘, TAL2 Probe:
5⬘-FAM-CTCTTCCCTCAAGGACCCCACCTGC-best distinguish among cases expressing known T-All onco-TAMRA-3⬘, LYL1 FW: CCCACTTTGGCCCTGCA-3⬘, LYL1 RV: 5⬘-GGTCCTGCTGGCCCAATGT3⬘, LYL1 Probe: 5⬘-FAM-TACCACCCTCACC
genes. This approach was based on the hypothesis that
domi-CCTTCCTCAACAGTGTC-TAMRA-3⬘ BHLHB1 FW: 5⬘-GGCAGTGGCTT
nant oncogenic transcription factors in this disease, such as
CAAGTCGTC-3⬘, BHLHB1 RV: 5⬘-TCCGGCTCTGTCATTTGCTT-3⬘,
HOX11, TAL1, and LYL1, stand at the top of regulatory cascades
BHLHB1 Probe: 5⬘-FAM-TCGTCCAGCACCTCGTCGTCTACG-TAMRA-3⬘,
whose aberrant activation can lead to T cell neoplasia. Our
HOX11 FW: 5⬘-TGGATGGAGAGTAACCGCAGAT-3⬘, HOX11 RV:
5⬘-“hypothesis driven” approach to hierarchical clustering has en-GGGCGTCCGGTTCTGATA-3⬘, HOX11 Probe: 5⬘-FAM-CACAAAGGACAG
abled us to integrate complex gene expression patterns into aGTTCACAGGTCACCC-TAMRA-3⬘, HOX11L1 FW: 5⬘-GGATGCTGGGTC CACACAAC-3⬘, HOX11L1 RV: 5⬘-CAGGATCTGATCGATGCCGA-3⬘,
conceptual framework with biologic relevance to T cell ALL. We
HOX11L1 Probe: 5⬘-FAM-TCCCACACCACGAGCCAATCAGC-TAMRA-3⬘, Acknowledgments
HOX11L2 FW: 5⬘-GCCCAAGCGTAAGAAGCCGC-3⬘, HOX11L2 RV: 5⬘-AGC
We thank Pablo Tamayo for help with the microarray analysis, Michael GCTTTTCCAGCTCGCAG-3⬘, HOX11L2 Probe: 5⬘-FAM-CACGTCCTTTTC
Hancock and Yinmei Zhou for help with the prognostic analysis, David CCGGGTGCAGA-TAMRA-3⬘, LMO1 FW: 5⬘-TCTACACCAAGGCCAACC
Zahrieh for assistance with the hierarchical dendrograms, John Gilbert for TCA-3⬘, LMO1 Probe:
5⬘-FAM-CGCGACTACCTGAGGCTCTTTGGCA-editorial review and substantive comments, and Craig Bassing for critical TAMRA-3⬘, LMO1 RV:TGCAAGCAGCACAGTTCCC-3⬘, LMO2 FW:
5⬘-review of the manuscript. This work was supported by NIH grants CA 68484 TACAAACTGGGCCGGAAGC-3⬘, LMO2 Probe: 5⬘-FAM-CGGAGAGACTAT
and CA 21765, the American Lebanese Syrian Associated Charities (ALSAC), CTCAGGCTTTTTGGGC-TAMRA-3⬘, LMO2 RV:5⬘-CTTGTCACAGGATGCG
St. Jude Children’s Research Hospital, Bristol-Myers Squibb, Millennium CAGA-3⬘. Unmodified primers and 5⬘-FAM, 3⬘-TAMRA or 5⬘-HEX, 3⬘-TAMRA
Pharmaceuticals, and Affymetrix, Inc. A.A.F. is a Fellow of the Leukemia and labeled probes were synthesized by Integrated DNA Technologies, while
Lymphoma Society. GAPDH 5⬘-VIC, 3⬘-TAMRA labeled probe was synthesized by PE Applied
Biosystems. Tal1d, FLT3 ITD analysis by PCR and MLL-ENL RT-PCR fusion transcript detection were performed as previously described (Meshinchi et al., 2001; Pongers-Willemse et al., 1999; Rubnitz et al., 1996).
Received: December 28, 2001 Revised: January 17, 2002
Real-time quantitative RT-PCR analyses were performed with an ABI PRISM References 7700 Sequence Detection System instrument (Perkin-Elmer Applied
Biosys-tems) in a total volume of 50l containing 1⫻ Taq DNA polymerase buffer
Allzadeh, A.A., Elsen, M.B., Davis, R.E., Ma, C., Lossos, I.S., Rosenwald, (Perkin-Elmer Applied Biosystems) supplemented with 1M 5-carboxy-X- A., Boldrick, J.C., Sabet, H., Ran, T., Yu, X., et al. (2000). Distinct types of Rhodamine (5-ROX) (Molecular Probes), 200M dATP, dCTP, dGTP, and diffuse large B-cell lymphoma identified by gene expression profiling. Nature dTTP (Roche), 100–300 nM of each primer, 150 nM of the taqman probe, 403, 503–511.
10 units of RNase inhibitor (Perkin-Elmer Applied Biosystems), 5 units of
Aplan, P.D., Lombardi, D.P., Ginsberg, A.M., Cossman, J., Bertness, V.L., reverse transcriptase (Perkin-Elmer Applied Biosystems), 1.25 units of
and Kirsch, I.R. (1990). Disruption of the human SCL locus by “illegitimate” AmpliTaq Gold DNA polymerase (Perkin-Elmer Applied Biosystems), and V-(D)-J recombinase activity. Science 250, 1426–1429.
1.5–2.5 mM MgCl2. One-step RT-PCR consisted of an initial incubation at
Aplan, P.D., Lombardi, D.P., and Kirsch, I.R. (1991). Structural characteriza-48⬚C for 30 min, followed by a denaturation step at 95⬚C for 10 min, and
tion of SIL, a gene frequently disrupted in T-cell acute lymphoblastic leuke-amplification for 40 cycles of 15 s at 95⬚C and 1 min at the optimized
mia. Mol. Cell. Biol. 11, 5462–5469. annealing temperature. 100 ng of total RNA was analyzed for each case.
RNA samples were treated before analysis with DNase I in order to eliminate Aplan, P.D., Lombardi, D.P., Reaman, G.H., Sather, H.N., Hammond, G.D., residual genomic DNA. Each experiment included three nontemplate con- and Kirsch, I.R. (1992). Involvement of the putative hematopoietic transcrip-tion factor SCL in T-cell acute lymphoblastic leukemia. Blood 79, 1327–1333. trols to detect any template contamination; a control lacking reverse
tran-scriptase was included for each sample to detect any residual genomic
Baer, R. (1993). TAL1, TAL2, and LYL1: a family of basic helix-loop-helix DNA. Standard curves were constructed by serial 10-fold dilutions from 108
proteins implicated in T cell acute leukaemia. Sem. Cancer Biol. 4, 341–347. to 102copies of in vitro-transcribed RNA of each gene to be quantified. The
Bain, G., Engel, I., Maandag, E.C., te Riele, H.P.J., Voland, J.R., Sharp, L.L., reproducibility of the quantitative measurements was evaluated by
conduct-Chun, J., Huey, B., Pinkel, D., and Murre, C. (1997). E2A deficiency leads ing triplicate PCR assessments. Glyceraldehyde-3-phosphate
dehydroge-to abnormalities in␣␤ T-cell development and to rapid development of T-cell nase (GAPDH) mRNA expression was analyzed for each sample in parallel
lymphomas. Mol. Cell. Biol. 17, 4782–4891. wells to assess the integrity of the RNA. INK4A-ARF deletion status was
determined by multiplex real-time quantitative PCR of INK4A/ARF and Bash, R.O., Hall, S., Timmons, C.F., Crist, W.M., Amylon, M., Smith, R.G., and Baer, R. (1995). Does activation of the TAL1 gene occur in a majority ␤-actin as described previously (Carter et al., 2001).
of patients with T-cell acute lymphoblastic leukemia? A Pediatric Oncology Group study. Blood 86, 666–676.
Microarray expression analysis
A total of 10 g of total RNA from each sample was used to prepare Begley, C.G., and Green, A.R. (1999). The SCL gene: from case report to critical hematopoietic regulator. Blood 93, 2760–2770.
biotinilated target RNA as previously described with minor modifications (Golub et al., 1999; Tamayo et al., 1999; Wodicka et al., 1997). A complete
Begley, C.G., Aplan, P.D., Davey, M.P., Nakahara, K., Tchorz, K., Kurtzberg, description of biochemical and analytical procedures is available as
Sup-J., Hershfield, M.S., Haynes, B.F., Cohen, D.I., Waldmann, T.A., et al. (1989). plemental Data at http://www.genome.wi.mit.edu/mpr and http://www. Chromosomal translocation in a human leukemic stem-cell line disrupts the cancercell.org/cgi/content/full/1/1/75/DC1. To identify the genes whose ex- T-cell antigen receptor delta-chain diversity region and results in a previously pression patterns best distinguished among HOX11⫹, TAL1⫹, LYL1⫹, and unreported fusion transcript. Proc. Natl. Acad. Sci. USA 86, 2031–2035. other cases, we analyzed the permutation distribution of the maximum t
Behm, F.G., Raimondi, S.C., Frestedt, J.L., Liu, Q., Crist, W.M., Downing, statistic using the Permax program (biowww.dfci.harvard.edu/gray/permax.
J.R., Rivera, G.K., Kersey, J.H., and Pui, C.H. (1996). Rearrangement of html). Affymetrix average difference values were transformed to base two
the MLL gene confers a poor prognosis in childhood acute lymphoblastic logarithms after raw values under 100 were raised to this value. 72 genes leukemia, regardless of presenting age. Blood 87, 2870–2877.
with a P value of⬍0.30 by permutation test analysis were used to build
Bene, M.C., Castoldi, G., Knapp, W., Ludwig, W.D., Matutes, E., Orfao, A., a hierarchical tree (http://www.insightful.com/products/splus). We chose a
and van’t Veer, M.B. (1995). Proposals for the immunological classification cutoff Permax P value of⬍0.30, even though P values ⬍0.05 are typically
of acute leukemias. European Group for the Immunological Characterization regarded as statistically significant. In our analysis, the Permax P value is
of Leukemias (EGIL). Leukemia 9, 1783–1786. calculated by comparing observed t statistics to the permutation distribution
of the largest t statistic obtained for the 5990 genes. Thus, a Permax P value Bernard, O., Guglielmi, P., Jonveaux, P., Cherif, D., Gisselbrecht, S., Mau-chauffe, M., Berger, R., Larsen, C.J., and Mathieu-Mahul, D. (1990). Two of 0.30 corresponds approximately to a t statistic of 4.73 with 17 degrees of
distinct mechanisms for the SCL gene activation in the t(1;14) translocation freedom, which has a nominal P value of 0.000097.
of T-cell leukemias. Genes Chromosomes Cancer 1, 194–208. To test whether the structure of the dendrogram shown in Figure 4
might be highly dependent on the limited number of genes used or our Bernard, O., Lecointe, N., Jonveaux, P., Souyri, M., Mauchauffe, M., Berger, method of gene selection, we increased the number to 116 genes selected R., Larsen, C.J., and Mathieu-Mahul, D. (1991). Two site-specific deletions based on a P value⬍0.5 instead of ⬍0.3, and obtained a nearly identical and t(1;14) translocation restricted to human T-cell acute leukemias disrupt
the 5⬘ part of the tal-1 gene. Oncogene 6, 1477–1488. dendrogram.
Bernard, O.A., Busson-LeConiat, M., Ballerini, P., Mauchauffe, M., Della, V., lar classification of cancer: class discovery and class prediction by gene expression monitoring. Science 286, 531–537.
Monni, R., Nguyen, K.F., Mercher, T., Penard-Lacronique, V., Pasturaud, P., et al. (2001). A new recurrent and specific cryptic translocation,
Greenberg, J.M., Boehm, T., Sofroniew, M.V., Keynes, R.J., Barton, S.C., t(5;14)(q35;q32), is associated with expression of the Hox11L2 gene in T
Norris, M.L., Surani, M.A., Spillantini, M.G., and Rabbits, T.H. (1990). Seg-acute lymphoblastic leukemia. Leukemia 15, 1495–1504.
mental and developmental regulation of a presumptive T-cell oncogene in the central nervous system. Nature 344, 158–160.
Boehm, T., Foroni, L., Kaneko, Y., Perutz, M.F., and Rabbitts, T.H. (1991). The rhombotin family of cysteine-rich LIM-domain oncogenes: distinct
mem-Griffiths, E.K., Krawczyk, C., Kong, Y.Y., Raab, M., Hyduk, S.J., Bouchard, bers are involved in T-cell translocations to human chromosomes 11p15
D., Chan, V.S., Kozieradzki, I., Oliveira-Dos-Santos, A.J., Wakeham, A., et and 11p13. Proc. Natl. Acad. Sci. USA 88, 4367–4371.
al. (2001). Positive regulation of T cell activation and integrin adhesion by the adapter Fyb/Slap. Science 293, 2260–2263.
Brown, L., Cheng, J.T., Chen, Q., Siciliano, M.J., Crist, W., Buchanan, G., and
Baer, R. (1990). Site-specific recombination of the tal-1 gene is a common Hatano, M., Roberts, C.W., Minden, M., Crist, W.M., and Korsmeyer, S.J. occurrence in human T cell leukemia. EMBO J. 9, 3343–3351. (1991). Deregulation of a homeobox gene, HOX11, by the t(10;14) in T cell
leukemia. Science 253, 79–82. Carter, T.L., Watt, P.M., Kumar, R., Burton, P.R., Reaman, G.H., Sather,
H.N., Baker, D.L., and Kees, U.R. (2001). Hemizygous p16INK4Adeletion in
Hawley, R.G., Fong, A.Z., Reis, M.D., Zhang, N., Lu, M., and Hawley, T.S. pediatric acute lymphoblastic leukemia predicts independent risk of relapse. (1997). Transforming function of the HOX11/TCL3 homeobox gene. Cancer
Blood 97, 572–574. Res. 57, 337–345.
Chen, Q., Cheng, J.T., Tasi, L.H., Schneider, N., Buchanan, G., Carroll, A., Herblot, S., Steff, A.M., Hugo, P., Aplan, P.D., and Hoang, T. (2000). SCL Crist, W., Ozanne, B., Siciliano, M.J., and Baer, R. (1990). The tal gene and LMO1 alter thymocyte differentiation: inhibition of E2A-HEB function undergoes chromosome translocation in T cell leukemia and potentially and pre-Talpha chain expression. Nat. Immunol. 1, 138–144.
encodes a helix-loop-helix protein. EMBO J. 9, 415–424.
Hsu, H.L., Wadman, I., and Baer, R. (1994). Formation of in vivo complexes Chervinsky, D.S., Zhao, X.-F., Lam, D.U., Ellsworth, M.K., Gross, K.W., and between the TAL1 and E2A polypeptides of leukemic T cells. Proc. Natl. Aplan, P.D. (1999). Disordered T-cell development and T-cell malignancies Acad. Sci. USA 91, 3181–3185.
in SCL LMO1 double-transgenic mice: parallels with E2A-deficient mice.
Kawabe, T., Muslin, A.J., and Korsmeyer, S.J. (1997). HOX11 interacts with Mol. Cell. Biol. 19, 5025–5035.
protein phosphatases PP2A and PP1 and disrupts a G2/M cell-cycle check-Chessells, J.M., Bailey, C., and Richards, S.M. (1995). Intensification of point. Nature 385, 454–458.
treatment and survival in all children with lymphoblastic leukaemia: results
Keller, G., Wall, C., Fong, A.Z., Hawley, T.S., and Hawley, R.G. (1998). of UK Medical Research Council trial UKALL X. Medical Research Council
Overexpression of HOX11 leads to immortalization of embryonic precursors Working Party on Childhood Leukaemia. Lancet 345, 143–148.
with both primitive and definitive hematopoietic potential. Blood 92, 877–887. Condorelli, G.I., Facchiano, F., Valtieri, M., Proietti, E., Vitelli, I., Huebner,
Kelliher, M.A., Seldin, D.C., and Leder, P. (1996). TAL-1 induces T cell acute K., Peschle, C., and Croce, C.M. (1996). T-cell-directed TAL-1 expression
lymphoblastic leukemia accelerated by casein kinase II␣. EMBO J. 15, 5160– induces T-cell malignancies in transgenic mice. Cancer Res. 56, 5113–5119.
5166. Dear, T.N., Colledge, W.H., Carlton, M.B., Lavenir, I., Larson, T., Smith, A.J.,
Kennedy, M.A., Gonzalez-Sarmiento, R., Kees, U.R., Lampert, F., Dear, N., Warren, A.J., Evans, M.J., Sofroniew, M.V., and Rabbitts, T.H. (1995). The
Boehm, T., and Rabbitts, T.H. (1991). HOX11, a homeobox-containing T-cell Hox11 gene is essential for cell survival during spleen development.
Develop-oncogene on human chromosome 10q24. Proc. Natl. Acad. Sci. USA 88, ment 121, 2909–2915.
8900–8904. Drexler, H.G. (1998). Review of alterations of the cyclin-dependent kinase
Larson, R.C., Fisch, P., Larson, T.A., Lavenir, I., Langford, T., King, G., inhibitor INK4 family genes p15, p16, p18 and p19 in human
leukemia-and Rabbitts, T.H. (1994). T cell tumours of disparate phenotype in mice lymphoma cells. Leukemia 12, 845–859.
transgenic for Rbtn-2. Oncogene 9, 3675–3681. Dube, I.D., Kamel-Reid, S., Yuan, C.C., Lu, M., Wu, X., Corpus, G., Raimondi,
Larson, R.C., Lavenir, I., Larson, T.A., Baer, R., Warren, A.J., Wadman, I., S.C., Crist, W.M., Carroll, A.J., Minowada, J., and Baker, J.B. (1991). A novel
Nottage, K., and Rabbitts, T.H. (1996). Protein dimerization between Lmo2 human homeobox gene lies at the chromosome 10 breakpoint in lymphoid
(Rbtn2) and Tal1 alters thymocyte development and potentiates T cell tumori-neoplasias with chromosomal translocation t(10;14). Blood 78, 2996–3003.
genesis in transgenic mice. EMBO J. 15, 1021–1027. Engel, I., Johns, C., Bain, G., Rivera, R.R., and Murre, C. (2001). Early
Look, A.T. (1997). Oncogenic transcription factors in the human acute leuke-thymocyte development is regulated by modulation of E2A protein activity.
mias. Science 278, 1059–1064. J. Exp. Med. 194, 733–745.
Lu, M., Gong, Z.Y., Shen, W.F., and Ho, A.D. (1991). The tcl-3 proto-onco-Finger, L.R., Kagan, J., Christopher, G., Kurtzberg, J., Hershfield, M.S.,
gene altered by chromosomal translocation in T-cell leukemia codes for a Nowell, P.C., and Croce, C.M. (1989). Involvement of the TCL5 gene on
homeobox protein. EMBO J. 10, 2905–2910. human chromosome 1 in T-cell leukemia and melanoma. Proc. Natl. Acad.
Sci. USA 86, 5039–5043. Ma, A., Pena, J.C., Chang, B., Margosian, E., Davidson, L., Alt, F.W., and
Thompson, C.B. (1995). Bclx regulates the survival of double-positive thymo-Fitzgerald, T.J., Neale, G.A., Raimondi, S.C., and Goorha, R.M. (1991). c-tal,
cytes. Proc. Natl. Acad. Sci. USA 92, 4763–4767. a helix-loop-helix protein, is juxtaposed to the T-cell receptor-beta chain
gene by a reciprocal chromosomal translocation: t(1;7)(p32;q35). Blood 78, McGuire, E.A., Hockett, R.D., Pollock, K.M., Bartholdi, M.F., O’Brien, S.J.,
2686–2695. and Korsmeyer, S.J. (1989). The t(11;14)(p15;q11) in a T-cell acute
lympho-blastic leukemia cell line activates multiple transcripts, including ttg-1, a Galy, A., Verma, S., Barcena, A., and Spits, H. (1993). Precursors of
gene encoding a potential zinc finger protein. Mol. Cell. Biol. 9, 2124–2132. CD3⫹CD4⫹CD8⫹ cells in the human thymus are defined by expression of
CD34. Delineation of early events in human thymic development. J. Exp. McGuire, E.A., Rintoul, C.E., Sclar, G.M., and Korsmeyer, S.J. (1992). Thymic
Med. 178, 391–401. overexpression of Ttg-1 in transgenic mice results in T-cell acute
lymphob-lastic leukemia/lymphoma. Mol. Cell. Biol. 12, 4186–4196. Geng, L., Pfister, S., Kraeft, S.K., and Rudd, C.E. (2001). Adaptor FYB
(Fyn-binding protein) regulates integrin-mediated adhesion and mediator release: Mellentin, J.D., Smith, S.D., and Cleary, M.L. (1989). Lyl-l, A novel gene differential involvement of the FYB SH3 domain. Proc. Natl. Acad. Sci. USA altered by chromosomal translocation in T-cell leukemia, codes for a protein
98, 11527–11532. with a helix-loop-helix DNA binding motif. Cell 58, 77–83.
Golub, T.R., Slonim, D.K., Tamayo, P., Huard, C., Gassenbeek, M., Mesirov, Meshinchi, S., Woods, W.G., Stirewalt, D.L., Sweetser, D.A., Buckley, J.D., Tjoa, T.K., Bernstein, I.D., and Radich, J.P. (2001). Prevalence and prognostic J.P., Coller, H., Loh, M.L., Downing, J.R., Caligiuri, M.A., et al. (1999).
Molecu-significance of Flt3 internal tandem duplication in pediatric acute myeloid Rodewald, H.R., and Fehling, H.J. (1998). Molecular and cellular events in
leukemia. Blood 97, 89–94. early thymocyte development. Adv. Immunol. 69, 1–112.
Min, H., Turck, C.W., Nikolic, J.M., and Black, D.L. (1997). A new regulatory Royer-Pokora, B., Loos, U., and Ludwig, W.D. (1991). TTG-2, a new gene protein, KSRP, mediates exon inclusion through an intronic splicing en- encoding a cysteine-rich protein with the LIM motif, is overexpressed in hancer. Genes Dev. 11, 1023–1036. acute T-cell leukaemia with the t(11;14)(p13;q11). Oncogene 6, 1887–1893. Miyamoto, A., Cui, X., Naumovski, L., and Cleary, M.L. (1996). Helix-loop- Rubnitz, J.E., Behm, F.G., Curcio-Brint, A.M., Pinheiro, R.P., Carroll, A.J., helix proteins LYL1 and E2a form heterodimeric complexes with distinctive Raimondi, S.C., Shurtleff, S.A., and Downing, J.R. (1996). Molecular analysis DNA-binding properties in hematolymphoid cells. Mol. Cell. Biol. 16, 2394– of t(11;19) breakpoints in childhood acute leukemias. Blood 87, 4804–4808. 2401.
Rubnitz, J.E., Camitta, B.M., Mahmoud, H., Raimondi, S.C., Carroll, A.J., Murre, C. (2000). Intertwining proteins in thymocyte development and cancer. Borowitz, M.J., Shuster, J.J., Link, M.P., Pullen, D.J., Downing, J.R., et al.
Nat. Immunol. 1, 97–98. (1999a). Childhood acute lymphoblastic leukemia with the MLL-ENL fusion
and t(11;19)(q23;p13.3) translocation. J. Clin. Oncol. 17, 191–196. Nakao, M., Yokota, S., Iwai, T., Kaneko, H., Horiike, S., Kashima, K., Sonoda,
Y., Fujimoto, T., and Misawa, S. (1996). Internal tandem duplication of the Rubnitz, J.E., Camitta, B.M., Mahmoud, H., Raimondi, S.C., Carroll, A.J., flt3 gene found in acute myeloid leukemia. Leukemia 10, 1911–1918. Borowitz, M.J., Shuster, J.J., Link, M.P., Pullen, D.J., Downing, J.R., et al. (1999b). Childhood acute lymphoblastic leukemia with the MLL-ENL fusion Neale, G.A., Rehg, J.E., and Goorha, R.M. (1995). Ectopic expression of
and t(11;19)(q23;p13.3) translocation. J. Clin. Oncol. 17, 191–196. rhombotin-2 causes selective expansion of CD4-CD8-lymphocytes in the
thymus and T-cell tumors in transgenic mice. Blood 86, 3060–3071. Schrappe, M., Reiter, A., Ludwig, W.D., Harbott, J., Zimmermann, M., Hidde-mann, W., Niemeyer, C., Henze, G., Feldges, A., Zintl, F., et al. (2000). Okuda, T., Shurtleff, S.A., Valentine, M.B., Raimondi, S.C., Head, D.R., Behm,
Improved outcome in childhood acute lymphoblastic leukemia despite re-F., Curcio-Brint, A.M., Liu, Q., Pui, C.H., Sherr, C.J., et al. (1995). Frequent
duced use of anthracyclines and cranial radiotherapy: results of trial ALL-deletion of p16INK4a/MTS1 and p15INK4b/MTS2 in pediatric acute lymphoblastic
BFM 90. German-Austrian-Swiss ALL-BFM Study Group. Blood 95, 3310– leukemia. Blood 85, 2321–2330.
3322. Palker, T.J., Fong, A.M., Scearce, R.M., Patel, D.D., and Haynes, B.F. (1998).
Shanahan, F., Seghezzi, W., Parry, D., Mahony, D., and Lees, E. (1999). Developmental regulation of lymphocyte-specific protein 1 (LSP1)
expres-Cyclin E associates with BAF155 and BRG1, components of the mammalian sion in thymus during human T-cell maturation. Hybridoma 17, 497–507.
SWI-SNF complex, and alters the ability of BRG1 to induce growth arrest. Mol. Cell. Biol. 19, 1460–1469.
Park, S.T., and Sun, X.H. (1998). The Tal1 oncoprotein inhibits E47-mediated transcription. Mechanism of inhibition. J. Biol. Chem. 273, 7030–7037.
Shirasawa, S., Arata, A., Onimaru, H., Roth, K.A., Brown, G.A., Horning, S., Arata, S., Okumura, K., Sasazuki, T., and Korsmeyer, S.J. (2000). Rnx Peterson, E.J., Woods, M.L., Dmowski, S.A., Derimanov, G., Jordan, M.S.,
deficiency results in congenital central hypoventilation. Nat. Genet. 24, 287– Wu, J.N., Myung, P.S., Liu, Q.H., Pribila, J.T., Freedman, B.D., et al. (2001).
290. Coupling of the TCR to integrin activation by Slap-130/Fyb. Science 293, 2263–2265.
Silverman, L.B., Gelber, R.D., Dalton, V.K., Asselin, B.L., Barr, R.D., Clavell, Pilarski, L.M., Turley, E.A., Shaw, A.R., Gallatin, W.M., Laderoute, M.P., L.A., Hurwitz, C.A., Moghrabi, A., Samson, Y., Schorin, M.A., et al. (2001). Gillitzer, R., Beckman, I.G., and Zola, H. (1991). FMC46, a cell protrusion- Improved outcome for children with acute lymphoblastic leukemia: results associated leukocyte adhesion molecule-1 epitope on human lymphocytes of Dana-Farber Consortium Protocol 91–01. Blood 97, 1211–1218. and thymocytes. J. Immunol. 147, 136–143.
Tamayo, P., Slonim, D., Mesirov, J., Zhu, Q., Kitareewan, S., Dmitrovsky, E., Pongers-Willemse, M.J., Seriu, T., Stolz, F., d’Aniello, E., Gameiro, P., Pisa, Lander, E.S., and Golub, T.R. (1999). Interpreting patterns of gene expression P., Gonzalez, M., Bartram, C.R., Panzer-Grumayer, E.R., Biondi, A., et al. with self-organizing maps: methods and application to hematopoietic differ-(1999). Primers and protocols for standardized detection of minimal residual entiation. Proc. Natl. Acad. Sci. USA 96, 2907–2912.
disease in acute lymphoblastic leukemia using immunoglobulin and T cell
Terstappen, L.W., Huang, S., and Picker, L.J. (1992). Flow cytometric assess-receptor gene rearrangements and TAL1 deletions as PCR targets: report
ment of human T-cell differentiation in thymus and bone marrow. Blood 79, of the BIOMED-1 CONCERTED ACTION: investigation of minimal residual
666–677. disease in acute leukemia. Leukemia 13, 110–118.
Terszowski, G., Jankowski, A., Hendriks, W.J., Rolink, A.G., and Kisielow, Pui, C.H., and Evans, W.E. (1998). Acute lymphoblastic leukemia. N. Engl.
P. (2001). Within the hemopoietic system, LAR phosphatase is a T cell J. Med. 33, 605–615.
lineage-specific adhesion receptor-like protein whose phosphatase activity Pullen, J., Shuster, J.J., Link, M., Borowitz, M., Amylon, M., Carroll, D., Land, appears dispensable for T cell development, repertoire selection and func-V., Look, A.T., McIntyre, B., and Camitta, B. (1999). Significance of commonly tion. Eur. J. Immunol. 31, 832–840.
used prognostic factors differs for children with T-cell acute lymphocytic
Tomayko, M.M., Punt, J.A., Bolcavage, J.M., Levy, S.L., Allman, D.M., and leukemia (ALL), as compared to those with B-precursor ALL: a Pediatric
Cancro, M.P. (1999). Expression of the Bcl-2 family member A1 is develop-Oncology Group (POG) study. Leukemia 13, 1696–1707.
mentally regulated in T cells. Int. Immunol. 11, 1753–1761. Reed, J.C. (1995). Regulation of apoptosis by bcl-2 family proteins and its
Vacchio, M.S., Ashwell, J.D., and King, L.B. (1998). A positive role for thymus-role in cancer and chemoresistance. Curr. Opin. Oncol. 7, 541–546.
derived steroids in formation of the T-cell repertoire. Ann. N Y Acad. Sci. Reinherz, E.L., and Schlossman, S.F. (1980). The differentiation and function 840, 317–327.
of human T lymphocytes. Cell 19, 821–827.
Valge-Archer, V.E., Osada, H., Warren, A.J., Forster, A., Li, J., Baer, R., and Rivera, G.K., Raimondi, S.C., Hancock, M.L., Behm, F.G., Pui, C.-H., Abro- Rabbitts, T.H. (1994). The LIM protein RBTN2 and the basic helix-loop-helix mowitch, M., Mirro, J., Jr., Ochs, J.S., Look, A.T., Williams, D.L., et al. protein TAL1 are present in a complex in erythroid cells. Proc. Natl. Acad. (1991). Improved outcome in childhood acute lymphoblastic leukaemia with Sci. USA 91, 8617–8621.
reinforced early treatment and rotational combination chemotherapy. Lancet
Veis, D.J., Sentman, C.L., Bach, E.A., and Korsmeyer, S.J. (1993). Expression
of the Bcl-2 protein in murine and human thymocytes and in peripheral T Roberts, C.W.M., Shutter, J.R., and Korsmeyer, S.J. (1994). Hox11 controls lymphocytes. J. Immunol. 151, 2546–2554.
the genesis of the spleen. Nature 368, 747–749.
Versteege, I., Sevenet, N., Lange, J., Rousseau-Merck, M.F., Ambros, P., Roberts, C.W., Galusha, S.A., McMenamin, M.E., Fletcher, C.D., and Orkin, Handgretinger, R., Aurias, A., and Delattre, O. (1998). Truncating mutations S.H. (2000). Haploinsufficiency of Snf5 (integrase interactor 1) predisposes of hSNF5/INI1 in aggressive paediatric cancer. Nature 394, 203–206. to malignant rhabdoid tumors in mice. Proc. Natl. Acad. Sci. USA 97, 13796–